GPT-5.4 and Gemini 3.1: OpenAI and Google Redefine AI

OpenAI Revolutionizes with GPT-5.4 Pro

OpenAI has recently launched the GPT-5.4 Pro version, marking a significant advancement in the field of artificial intelligence. This new version stands out with an impressive context window of 1 million tokens, allowing for deeper and more nuanced understanding and responses. Among other improvements, there is a mid-response course correction feature and an enhanced ability to use tools natively on a computer. The model's performance, measured by the GPT-VAL score, has reached 83%, demonstrating its advanced capabilities. Additionally, OpenAI has implemented strengthened security measures to ensure high cyber capacity.

In parallel, OpenAI has also introduced GPT-5.3 Instant, characterized by a less moralizing tone and a 26.8% reduction in hallucinations, thereby improving the user experience.

Google Enhances Gemini 3.1 Flash Lite

Google is not lagging behind with the update of Gemini 3.1 Flash Lite. This version offers faster access time to the first token and improved throughput, optimizing the efficiency of interactions. Google has also introduced a command-line interface, facilitating the integration of agents with services such as Gmail, Drive, and Docs. However, discussions have emerged regarding the potential risks of agent failures in real-world situations, illustrated by an incident involving mass email deletions by an AI.

Luma and Multimodal Innovation

Luma has launched unified multimodal models, enabling end-to-end content creation across various media such as text, image, video, and audio. A notable example of their effectiveness is an advertising localization project completed in just 40 hours for a cost of under $20,000.

Controversies and Challenges in the AI Sector

The AI sector is not without its controversies. Anthropic has been identified as a potential risk to the supply chain, leading to restrictions. OpenAI, for its part, has emphasized in its contracts with the DoD the importance of adhering to "all legal uses." Consumer cancellations have paradoxically boosted the ranking of Claude's applications, while OpenAI announced a fundraising of $110 billion, reaching an impressive valuation of $730 billion.

Additionally, Alibaba has lost key leaders from its Qwen project, and a lawsuit has implicated Gemini in a suicide case. Anthropic has also warned of a major disruption in the labor market, and METR has revised its estimates regarding the timeline for AI.